Fatigue test data in finite life region are normally analyzed separately for each stress level because of varying scatters, while another method assuming equal scatters on all levels and thus unifying all data into one model is also welcome for its efficient usage of data. Combining the idea of the two methods, a new method that assumes linear change of scatter according to stress levels is developed in this paper. The algorithm derives from maximum likelihood estimation and general Newton's method. Monte Carlo method is used for theoretical validation and the new method is proved to be correct with the estimations. Finally a set of test data which shows a tendency of trapezoidal data distribution is analyzed and an S-N curve for finite life range is created. The new method has its advantage in certain situations so it's meaningful to add it to the alternatives of fatigue data analysis methods.